Analysis of monocyte data (SCZ vs CTR): complex model

Load data

First step QC

Plotting the first QC results:

  1. Density plot

  2. Covariate correlation

  1. PC-covariate correlation

The model we use for the downstream analyses (DEA & data correction) is:
batch + RIN + percent.ERCC + stimulation

Next step: QC on vst-normalized expression values.

  1. Gene expression variance (low variance in lowly expressed genes (to the left) = good)

  1. PCA

Data correction

We next correct the data with the same model we use for the DESeq2 DEA (for plotting purposes post DEA).

The results are as follows: (1) PC-covariate correlation post-correction:

  1. PCA post-correction

DESeq2 anaylsis

We then go into the differential gene-expression analysis:

  1. Overview of the results

out of 35316 with nonzero total read count adjusted p-value < 0.1 LFC > 0 (up) : 812, 2.3% LFC < 0 (down) : 1257, 3.6% outliers [1] : 0, 0% low counts [2] : 0, 0% (mean count < 0) [1] see ‘cooksCutoff’ argument of ?results [2] see ‘independentFiltering’ argument of ?results

As well as the number of differentially expressed genes at lfc </> -1/1 at padj < 0.05:

[1] 1807

Interactive results table

  1. Results

Enrichment testing results

  1. List enrichments

[1] “Reminder: we have 787 upregulated and 1234 downregulated genes.”

  1. GO enrichments